Team:Duesseldorf/Demonstrate

Demonstration of our work

For creating synthetic milk, we took it upon ourselves to create all needed components without cows, using the power of synthetic biology. These include fatty acids, which are necessary for synthesizing milk´s lipids, and the proteins. For this, we created several subgroups and we can proudly tell that we were successful in every subgroup.

Fatty acids

The first central component of cow’s milk we worked on is the basic building block of lipids - the fatty acids. The central fatty acid metabolism is shown in Fig. 1. By modeling our subproject, we were able to select different candidate genes to either overexpress or repress in order to increase fatty acid yield. We therefore chose three separate approaches to maximize production in our various engineered strains, as well as efficiently quantify our product:

  1. Regulation of fatty acid chain length by expression of various chain-length-specific thioesterases
  2. Use of a CRISPR-dCas9 system to repress genes encoding proteins involved in either different steps of fatty acid synthesis or in undesirable, off-target reactions
  3. Development of a biosensor system for in vivo quantification of fatty acids produced by Escherichia coli production strains.

Fig. 1: Pathway of the fatty acid metabolism. The pathway was derived from Synechocystis sp. PCC  6803. This Figure shows genes and enzymes involved in the central fatty acid biosynthesis pathway. Target genes selected for repression are shown with red arrows, target genes for overexpression are shown with green arrows.
Modular fatty acid chain length regulation

For synthesizing a broad range of fatty acids, several thioesterases of different organisms were heterologously expressed in two different organisms - E. coli and Synechocystis1. A schematic view of our gene constructs is shown in Fig. 2.

Fig. 2: Scheme of construct. The insert, containing the promoter PJ23119 (BBa_J23119), RBS* (BBa_K2924009), a thioesterase gene (BBa_K2924001, BBa_K2924002, BBa_K2924003, BBa_K2924004) and the double terminator (BBa_B0015), was cloned via Gibson Cloning into the pSHDY or pSNDY backbone, which have a broad host range, enabling the conjugative transfer from E. coli to cyanobacteria and other microorganisms. The plasmid contains the standard BioBrick restriction sites EcoRI (E), XbaI (X), SpeI (S), PstI (P).

Results

To analyze the change in fatty acid yield of both organisms, we analyzed the fatty acid content using gas chromatography-mass spectrometry (GC-MS).
E. coli cultures were grown overnight at 37 °C and shaken at 250 rpm. The following day, 4 OD units (ODU) of cells were harvested (i.e. an equivalent of 1.2 ml cells at OD600 = 3.3). They were centrifuged at 4500 rpm for 15 minutes and the pellet was used for the extraction and derivatization of fatty acids.
Synechocystis cultures were grown under specific light conditions with 0.5 % CO2 at 30 °C and shaken at 150 rpm. These cells were also harvested at an ODU of 4 and centrifuged for further extraction and derivatization.
The rhamnose-inducible, leaderless thioesterase of E. coli, ’TesA (BBa_K1472601), was additionally induced with 0.5 mM rhamnose 24 h before harvesting.

Fig. 3: Relative fatty acid yield of transformants with the leaderless ‘TesA compared to the control strains. Different fatty acids are listed for E. coli (blue and red) and for Synechocystis (green and orange). Measurements were carried out via GC-MS.
As seen in Fig. 3, the E. coli transformants showed a higher fatty acid yield in the C16:1, C18:1 and in C20:1 fatty acids, with the highest fold change, compared to the E. coli control.
The Synechocystis conjugants showed a higher fatty acid yield in the C18:0 and C20:2 fatty acids, but a decrease in C16:0, C16:1 compared to the Synechocystis control, which is the same as the wild type (WT), was also detected.
Short- and medium-chain fatty acids were not present or could not be detected in all of the Synechocystis samples.
Fig. 4: Relative fatty acid yield of transformants with the thioesterase from Haematococcus pluvialis (TeHP) compared to the control strains. Different fatty acids are listed for E. coli (blue and red) and for Synechocystis (green and orange). Measurements were carried out via GC-MS.
As seen in Fig. 4, the E. coli transformants clearly showed a higher fatty acid yield in the C20:1 fatty acid compared to the control.
In contrast, the Synechocystis conjugants showed a slightly higher yield for the fatty acids C18:0 and C18:1, but in addition a slightly lower fatty acid yield in the fatty acids C16:1, C18:3 and 20:1.

To summarize, we were able to show that the fatty acid yield for a variety of chain lengths increased in our engineered strains, due to the overexpression or heterologous expression of the different thioesterases, such as ’tesA.

Use of CRISPRi/dCas9-system

Modeling of the fatty acid metabolism of Synechocystis showed promising candidates for down-regulation. Since we were unsure whether deletion of these genes might disrupt cell viability, we chose to try an alternative method - inducible down-regulation at the transcriptional level via a previously published CRISPRi/dCas9-system2, for increasing the yield of free fatty acids. The CRISPRi/dCas9-system was provided to us by Yao et al.2. In their system, both dCas9 and the gene-specific sgRNA are inducible with anhydrotetracycline (aTc). The dCas9 used is catalytically inactive and therefore does not have an endonuclease activity as the original Cas9. Therefore, this method is well-suited for down-regulating otherwise essential enzymes.
Since the system was already established for Synechocystis sp. PCC 6803, we used the background strain from Yao et al.2 encoding dCas9 and integrated a gene-specific small guide RNA (sgRNA) for each target, which is responsible for finding the targeted gene and recruiting dCas9, into the genome of Synechocystis sp. PCC 6803 to enable complex formation of dCas9 with the sgRNA and inhibit the transcription temporarily.

Fig. 5: Scheme of construct encoding a gene specific sgRNA. The inserts, which were synthesized by overlap extension PCR, containing the aTc-inducible promoter PL22, the specific sgRNA (orange), dCas9 binding site, terminator and kanamycin resistance cassette, were cloned via Gibson Cloning into the NS4 backbone, which enables the integration into the genome of Synechocystis sp. PCC 6803 via homologous recombination.

Results

Similar to the heterologous expression of thioesterases, the GC-MS method was also used for the knock-down targets. Synechocystis was grown under specific light and CO2 conditions at 30 °C and 150 rpm.
As proof of concept of the CRISPRi/dCas9 methodology, a sgRNA for mVenus was designed and coexpressed in a strain constitutively expressing mVenus (BBa_K2924036) and tested by measuring the fluorescence.
First, the temporal response was checked. Maximum reduction of fluorescence could be observed approximately 24 h after induction with aTc, as shown in Fig. 6.

Fig. 6: Effect of CRISPRi/dCas9-system on mVenus fluorescence over time. Measurements of fluorescence started with aTc induction at an OD750 of 0.4 in the plate reader and were carried out every 6 h. Prior to each measurement, an aliquot was sampled and the OD750 was adjusted to 0.4. Fluorescence was measured at λex/em = 511 / 552 nm. Blue and green symbols and lines denote two different biological replicates, each of which was measured in technical replicates.
Next, the fluorescence of the mVenus knock-down strain (BBa_K2924000 with mVenus-specific sgRNA) was cultured in the presence or absence of the inducer aTc prior to measuring fluorescence in the plate reader.
Fig. 7: Fluorescence measurement of the mVenus knock-down (KD) strain in the plate reader 24 h after induction with 500 nM aTc (red) or 100 % EtOH as a negative control (blue). 2 biological and 3 technical replicates were cultured in 6-well plates.
As seen in Fig. 7 the overall fluorescence of the induced Synechocystis knockdown strain decreased compared to the uninduced control strain. Nevertheless, the induced Synechocystis knockdown strain still showed some fluorescence compared to the empty vector control (EVC).

We then chose to further characterize the strains microscopically. For this purpose, Synechocystis sp. PCC 6803 with sgRNA_mVenus + pSHDY_Pcpc560_mVenus and Synechocystis sp. PCC 6803 with pSHDY_Pcpc560_mVenus cultures were diluted to an OD750 of 0.2 and induced with 500 nM aTc. As shown in Fig. 6 the most activity was detected after approximately 24 h. 1 ml each of induced and uninduced culture was sampled after exactly 24 h. These samples were further diluted 1:10 and used for fluorescence microscopy.
Fig. 8: Overview (A and B) and detail image (C and D) of the mVenus knock-down strain which were either induced with 100% EtOH (A and C) or 500 nM aTc (B and D), after 24 h of incubation. It shows clearly the chlorophyll autofluorescence (purple) and the mVenus fluorescent protein (green).
As shown in Fig. 7, when the culture is induced with 500 nM aTc, there is an overall lower fluorescence compared to the control culture, measured already in the plate reader. Interestingly, in contrast to our expectation, that this may be due to a dose-dependent, gradual decrease of mRNA, resulting in a gradual decrease of protein, it is more comparable to an “ON/OFF"-switch, which is seen in Fig. 8. Most of the induced cells show no evidence of mVenus fluorescence at all (Fig. 8 B and D), while a few remaining cells show the same fluorescence intensity compared to the control strain (Fig. 8 A and C). Also the localization of the chlorophyll in the membrane due to the autofluorescence, as well as the localization of the cytosolic mVenus fluorescent protein can be clearly seen in Fig. 8.

These results show that our basic concept of knocking down enzymes by the CRISPRi/dCas9-system2 was successful.

After proof-of-concept using mVenus, the aforementioned sgRNAs for enzymes involved in the fatty acid metabolism were designed and tested.

The long-chain-fatty-acid CoA ligase (aas, slr 1609) is responsible for ligating coenzyme A to a long-chain free fatty acid under a high consumption of energy. This step marks the initiation of the degradation of fatty acids. The enzyme was targeted and inhibited by the dCas9 via our gene-specific sgRNA (BBa_K2924005).
Several biological replicates ofSynechocystis sp. PCC 6803 cultures, encoding the specific sgRNA, were inoculated and grown under specific light and CO2 conditions at 30 °C and shaken at 150 rpm. The cultures were diluted to an OD750 of 0.2 and induced with 500 nM aTc.
Fig. 9: Relative fatty acid yield of Synechocystis sp. PCC 6803 transformants with the long-chain-fatty-acid CoA ligase-specific sgRNA compared to the control strains. Different fatty acids are listed and tested for two biological replicates. Measurements were carried out via GC-MS.
The results indicate a slight difference in the fatty acid profile and yield compared to the control strain (shown in Fig. 9). For the fatty acids C17:1, C18:0 and C18:1, there is a slight increase in fatty acid yield visible. This is likely due to the long-chain-fatty-acid CoA ligase knockdown.
These results depends on several effects. One effect that could influence the result is the efficiency of the designed sgRNA, which may bind or fold inefficiently, leading to weak specificity or binding properties. Currently, the efficiency of the sgRNA can only be experimentally determined3. Another aspect could be the light-mediated degradation of aTc, which results in a lower signal that can influence the expression, and therefore can influence the reduction of the enzyme.

Biosensor

Biosensors are an efficient, cheap method, especially useful when screening a large number of strains overproducing fatty acids. The production of long chain fatty acids (LCFA) is regulated by the transcription factor FadR4. This transcription factor can bind to Acyl-CoA which leads to the release of FadR from its cognate operator sequence, thereby increasing gene expression (as shown in Fig. 10).

Fig. 10: Regulation of the fatty acid metabolism by the transcriptional factor FadR. FadR recognizes its cognate binding site (white), thereby repressing transcription. Upon binding of an Acyl-CoA to FadR, the promoter region is freed, enabling gene expression.
We chose to test a synthetic fatty acid biosensor already present in the registry, comprising a double terminator the fatty acid inducible promoter PAR and the reporter gene for RFP5.
Since RFP has a relatively slow folding activity, we chose to improve this part by adding more reporter genes, namely, sfGFP (BBa_I746916), which has improved folding characteristics, and amilCP (BBa_K592009), which can be quantified by measuring absorption.
The different PAR biosensor variants were tested in a dose-dependent manner by adding different concentrations of the fatty acids with a chain length from C14:0 to C18:0 to the culture medium prior to culturing the cells. An EVC treated in the same manner was included in each measurement.
Fig. 11: Response of PARRFP (red) to different chain lengths of fatty acids compared to an empty vector control (gray). Plot A presents the response for lauric acid, plot B for myristic acid, plot C for palmitic acid and plot D for stearic acid.
By adding fatty acids to the culture medium, a dose dependent increase of fluorescence, consistent with the results previously shown by UPF_CRG_Barcelona iGEM team 2018 (BBa_K2581012), was expected, while the empty vector control (EVC) should remain at a basal level.
Fig. 11 shows that the addition of fatty acids to the culture medium leads to an increase in fluorescence in a dose-dependent fashion compared to the EVC. The most clear result could be shown with the fatty acids palmitic acid (C16:0) and stearic acid (C18:0). By adding myristic acid (C14:0) to the culture medium, fluorescence of the reporter gene RFP was increased, but the fluorescence of the EVC increased slightly as well. These results indicate that the promoter is more specific to the chain lengths C16:0 and C18:0 than to the chain lengths C12:0 and C14:0. The E. coli cells with the RFP were also monitored with confocal fluorescence microscopy to visualize the intracellular localization of RFP in the cells. The RFP appears to be located in the cytosol instead of being bound to a membrane (Fig. 12).
Fig. 12: Overview (A) and at detailed (B) image from RFP in a E. coli strain after being induced with 1 mM palmitic acid. The RFP can be detected in the whole cytosol.
Fig. 13: Response of PARsfGFP (green) to different chain lengths of fatty acids compared to an empty vector control (gray). Plot A presents the response for lauric acid, plot B for myristic acid, plot C for palmitic acid and plot D for stearic acid.
Fig. 13 shows the same dose-response experiments for the improved part, PAR:sfGFP. The graphs show a clearly increase in fluorescence at high fatty acid concentrations compared to the original part of Barcelona with RFP. Fluorescence is strongest in the experiment with stearic acid (C18:0), but in the other experiments a large increase in fluorescence can also be seen. For example, addition of the fatty acid palmitic acid (C16:0), led to a 2-fold increase of fluorescence from 0.01 mM to 1 mM could be observed. EVC also shows a better result. The fluorescence of the samples increased only minimally with increasing concentrations of fatty acids, with only a neglectable change in fluorescence compared to sfGFP, indicating some kind of excitation resulting directly from some of the fatty acids in the case of RFP-exciting wavelengths, instead of from the reporter protein. This is not the case for excitation wavelengths relevant for sfGFP, further supporting the improvement of our part. For induction with lauric acid (C12:0), the results appear more erratic at higher concentrations , which is probably due to a decreasing specificity of the promoter AR to the fatty acids - this result is also consistent with the data measured for PAR:RFP.
The E. coli cells expressing sfGFP were also monitored with confocal fluorescence microscopy to visualize the localization of sfGFP in the cells. As for RFP, sfGFP is also not bound to the membrane, but is localized in the cytosol (Fig. 14).
Fig. 14: Overview (A) and at detailed (B) image from sfGFP in a E. coli strain after being induced with 1 mM palmitic acid. The sfGFP can be detected in the whole cytosol.
Fig. 15: Response of PAR+amilCP (blue) to different chain lengths of fatty acids compared to an empty vector control (gray). Plot A presents the response for lauric acid, plot B for myristic acid, plot C for palmitic acid and plot D for stearic acid.
The same experiments using the blue chromoprotein amilCP (BBa_K592009) as a reporter showed that the production of the chromoprotein is increased by a higher concentration of fatty acids in the medium. The best result was achieved with the fatty acid stearic acid. The biosensor also worked for the chain lengths from C14:0 and C16:0. Here, the EVC is lower than the samples with amilCP. By adding lauric acid (C12:0) to the culture medium, the production of amilCP increased, but the EVC also increased in a similar manner close to amilCP, so it is unclear if this result offers relevant insights.

Another promising biosensor candidate we obtained from the literature is PaldA6. The promoter controlling aldA was published as an oleic acid (C18:1) sensitive promoter enabling for the measurement of free fatty acids present in microorganisms. In E. coli, the growth on fatty acids requires many different proteins which are repressed by the transcriptional factor FadR1. The long chain acyl-CoA ester is an effector on the transcriptional factor FadR for the regulation of fatty acid metabolism2. After adding oleic acid (C18:1) many proteins showed an altered expression level and new proteins like aldA were synthesised. The promoter for aldA was also used for promoting the synthesis of green fluorescent protein (GFP)6. We therefore isolated PaldA from the E. coli wild type genome and fused it to eYFP, as shown in Fig. 16.
Fig. 16: The promoter aldA is bounded to the reporter gene eYFP and cloned into a bBb backbone7. The restriction enzymes EcoRI and XbaI were used for the cloning. The pBb backbone has a kanamycin resistance and a medium copy ori p15A.
In the same manner as the experiments performed for PAR, we measured dose-dependent changes in fluorescence for different fatty acid inducers (Fig. 17).
Fig. 17: Response of PaldA:eYFP (yellow) to different chain lengths of fatty acids compared to an empty vector control (gray). Plot A presents the response for palmitic acid, plot B for stearic acid and plot C for oleic acid. The fluorescence was measured at an excitation wavelength from 497 nm and an emission wavelength from 540 nm.
The measurement showed that the PaldA promoter has an higher increase of eYFP when oleic acid (C18:1) was added to the culture. In addition, PaldA was also able to detect stearic acid (C18:0), the saturated form of oleic acid, as well as palmitic acid (C16:0), although with a weaker absolute fluorescence value than oleic acid (C18:1).

In vivo detection of fatty acids in our production strains

After testing the dose-dependent response of our biosensor constructs to fatty acids added extracellularly, we were interested in exploring the biosensor response to the intracellular increase of fatty acids via coexpression with heterologous genes in vivo. Production strains where we expected a higher production of long chain fatty acids were co-transformed with biosensor constructs for a fast measurement method to show which cultures can produce more fatty acids.
Fig. 18: Two thioesterases and a Acetyl-CoA carboxylase complex (ACC) were combined with biosensor PaldA:eYFP. The cultures were induced by 1 mM IPTG and 1 mM rhamnose (+) and the absolute fluorescence were measured. They were compared to the uninduced controls (-). eYFP fluorescence was excited at 497 nm and measured at an emission wavelength of 540 nm.
When cells expressing the Acetyl-CoA-Carboxylase (ACC) genes and a biosensor were combined and induced, the cells showed a slight increase of fatty acid production compared to the control (biosensor only). This increase was specific for the induced sample.
When the thioesterase ´TesA and the biosensor were combined and induced in the cells, the fluorescence was much higher than the control (biosensor only). Similar to ACC, fluorescence increased compared to the uninduced culture.

Finally, we chose to compare our previously obtained GC-MS results to our measurements with the biosensor. Fig. 19 shows the biosensor data compared to GC-MS data. We were able to show a strong correlation between the two methods. In summary, we could show that our biosensor not only works as a fast, efficient read-out for intracellular fatty acids, but that the data obtained from it corresponds to actual measured GC-MS data.
Fig. 19: Comparison of the in vivo biosensor PaldA (blue) to the results obtained via GC-MS (red) with stearic acid (18:0). The in vivo biosensor was diluted to an OD of 0.5 and induced by 1 mM IPTG and 1 mM rhamnose. For the GC, 4 ODU were used for the measurement.
Fig. 19 shows that the signal achieved by overexpressing ´TesA is clearly higher than in the wild type (WT) in both experiments. In the case of ACC, a slightly higher fold change could be detected compared to the WT. By comparing both methods to each other as shown in Fig. 19, we are able to show that our engineered system functions robustly.

Proteins

The second component of cow’s milk which we were interested in producing is the proteins. We focused on the six most abundant proteins present in cow’s milk, and applied various strategies to maximize efficiency. In order to provide a high degree of flexibility in the production process, various organisms - Escherichia coli, Bacillus subtilis, Pichia Pastoris and Synechocystis sp. PCC 6803 - were used to manufacture the desired milk proteins.



Escherichia coli

Fig. 20: Scheme of α-s1-casein overexpression construct. The insert, containing the promoter PT7 (BBa_K2406020), α-s1-casein gene (BBa_K2924026) fused with a 6x His-tag and the T7 terminator (BBa_B0012), was cloned into the pET22b backbone.
For proof of concept , the model organism E. coli (BL21) was used, because it carries the orthogonal, strong T7 polymerase in its genome and expresses fewer proteases that could degrade the product8. To heterologously produce the proteins of interest in a high yield, the strong T7 expression system, containing the PT7 promoter (BBa_K2406020) and T7 terminator (BBa_B0012), was used. α-s1-casein (BBa_K2924026) was cloned into pET22b and used to transform the production organism.
Fig. 21: SDS-PAGE of E. coli protein. The lysed cells of E. coli BL21 with pET22b+α-s1-casein and the empty vector control were loaded. The SDS-PAGE was run at 220 V, 400 A for 45 minutes and then stained with a Coomassie blue dye. α-s1-casein has an expected size of 25.4 kDa.
Fig. 21 shows a band of almost 25 kDa in the cell pellet sample of E. coli BL21 with α-s1-casein, there is no band visible in comparison to the empty vector control of E. coli. This band is very likely the α-s1-casein protein, which ran slightly lower than expected. The band is faint compared to most other protein bands on the gel, suggesting that the yield is low for this expression. Additionally, there is a second band visible at around 55 kDa, which is not present in the empty vector, which might be a dimer of α-s1-casein, since it has roughly twice the mass of a monomer.
The results indicate that E. coli could be suitable for the production of milk proteins, however, further experiments need to be carried out to observe the secretion efficiency for the protein and to determine how much of the protein can be produced.

Bacillus subtilis

After consulting Dr. Andreas Knapp from the Forschungszentrum Jülich, who provided us with extensive information on protein secretion, as well as relevant strains and plasmids, we decided to use B. subtilis in the hope of not only producing relevant amounts of protein, but being able to secrete them to the supernatant, thereby significantly facilitating downstream processing. The low protease Bacillus subtilis9, 10 strain was used as one of our main production organism to secrete the proteins into its culture medium11, 12, 13 with the help of diverse signal sequences. As there is no signal sequence that works for every protein, the only option was to try all possible options.

B. subtilis PHpaII + RBS + SPNprE+ α-s1-casein + 6xHis-tag + fd Terminator

This part contains the constitutive promoter PHpall (BBa_K2924043), expressing α-s1-casein (BBa_K2924026) with the secretion signal SPNprE (BBa_K2924047) and the fd terminator (BBa_K2924044) for expression of the protein in B. subtilis.

Fig. 22: Scheme of construct. The insert, containing the promoter PHpall (BBa_K2924043), α-s1-casein gene (BBa_K2924026) and the double fd terminator (BBa_K2924044), was cloned into the pBSMUl1 backbone with different secretion signals - here: SPNprE.
For expression and secretion of α-s1-casein, the gene was cloned into the high-copy pBSMUl1-SPNprE plasmid, N-terminally fused to the signal peptide SPNprE and C-terminally fused to a 6xHis-tag for easier purification and immunodetection.

Neither B. subtilis native nor heterologously expressed α-s1-casein secreted proteins from small-scale experiments were detectable on coomassie stained SDS-PAGE. To detect smaller amounts of protein, western blot analysis was performed on different B. subtilis fractions. α-s1-casein has a molecular weight of 25,4 kDa without the signal peptide and 28,5 kDa with the signal peptide. No bands could be detected in the SDS-PAGE gel for wild type and pBSMUl1-SPNprE+α-s1-casein. The expressed protein was also fused to a 6xHis tag. For detection, an anti-his antibody was used and further analyzed via Western Blot .
Fig. 23: Western Blot of Bacillus subtilis lysed pellet protein. The wild type B. subtilis DB430 shows no significant bands. pBSMUl1-SPNprE+α-s1-casein shows two bands at 15 kDa and ~25 kDa.
A western blot of the lysed cell pellet showed two intense bands for pBSMUl1-SPNprE+α-s1-casein at 15 kDa and ~25 kDa respectively (Fig. 23). The band at 25 kDa corresponds to the expected size of α-s1-casein, while the band at 15 kDa might be degraded protein.
To show that our protein of interest was secreted into the growth medium, another western blot was performed with the concentrated fraction of the supernatant, which should contain proteins over 30 kDa (Fig. 24). This fraction and not the fraction below 30 kDA was used, since the cutoff is close to the molecular size of the protein of interest and it therefore should be retained in this fraction.
Fig. 24: Western Blot of B. subtilis media fraction containing proteins around and over 30 kDa. Wild type = B. subtilis DB430. Both the wild type and pBSMu1+α-s1-casein show a faint band around ~25 kDa which have nearly the same intensity.
The western blot of the concentrated protein fraction containing proteins >30 kDa showed no difference between the wild type control and the strain expressing SpNpre+α-s1-casein, indicating that the protein was retained in the cells and not secreted into the medium (Fig. 24).

Conclusion

The gene α-s1-casein was successfully cloned and expressed in B. subtilis. With the focus on optimizing the expression of the protein of interest, five signal peptides (sslipA, SPNprE, SPPel, SPEpr and SPYurl) were tested. Western Blot analysis of lysed pellet proteins showed that the protein of interest remained inside the cells, indicating that the signal peptide SPNprE was suitable for α-s1-casein production (Fig. 23), but not for secretion, since no protein could be detected in the growth media or in the concentrated media fraction containing proteins >30 kDa (Fig. 24).

The band at 25 kDa resembles our protein of interest, while the band at 15 kDa might be degraded protein. After 12 h, the cells should have reached the stationary phase and do not produce new protein. The produced protein could be degraded after that time point. In further experiments, the induction time should be decreased to see if the unspecific band at 15 kDa is decreased and the specific protein band is increased.

B. subtilis Hpall + RBS + SPNprE + α-s2-casein + 6xHis-tag + fd terminator

This part contains the constitutive promoter PHpall (BBa_K2924043), expressing α-s2-casein (BBa_K2924027) with the secretion signal SPNprE (BBa_K2924047) and the fd terminator (BBa_K2924044) for expression and secretion of the protein in B. subtilis.
Fig. 25: Scheme of construct. The insert, containing the promoter PHpall (BBa_K2924043), α-s2-casein gene (BBa_K2924027) and the double fd terminator (BBa_K2924044), was cloned into the pBSMUl1 backbone with different secretion signals - here: SPNprE.
For expression and secretion of α-s2-casein, the gene was cloned into the high-copy pBSMUl1-SPNprE plasmid, N-terminally fused to the signal peptide SPNprE and C-terminally fused to a 6xHis-tag for easier purification and immunodetection.

Neither B. subtilis native nor heterologously expressed α-s2-casein secreted proteins from small-scale experiments were detectable on a Coomassie stained gel. To detect smaller amounts of protein, western blot analysis of different B. subtilis fractions was performed. α-s2-casein has a molecular weight of ~27 kDa without the secretion signal and ~30 kDa with the secretion peptide. No bands could be detected in the SDS-PAGE gel for WT and pBSMUl1-SPNprE+α-s2-casein. An anti-his antibody was used for Western Blot Analysis.

Fig. 26: Western Blot of B. subtilis lysed pellet protein. The wild type Bacillus subtilis DB430 shows no significant bands. pBSMUl1-SPNprE+α-s2-casein shows no significant bands but a faint band around 15 kDa.
α-s2-casein has a molecular weight of ~27 kDa without and ~30 kDa with the signal peptide. The western blot of the lysed pellet shows no strong band for the wild type, while showing a faint band for pBSMUl1-SPNprE+α-s2-casein at 15 kDa (Fig. 26), however there is no strong band at the expected sizes at ~27 or 30 kDa as there is for pBSMUl1-SPNprE+α-s1-casein. Another, more favorable, explanation could be, that the secretion tag works more efficiently for α-s2-casein then for α-s1-casein and therefore the protein is secreted more efficiently to the medium. To test this hypothesis, another western blot was performed with the concentrated fraction which should contain secreted proteins with a molecular weight above 30 kDa (Fig. 27). This fraction and not the fraction below 30 kDA was used, since the cutoff is close to the molecular size of the protein of interest and it therefore should be retained in this fraction.
Fig. 27: Western Blot of B. subtilis concentrated media fraction containing proteins around and over 30 kDa. Wild type= B. subtilis DB430. The wild type shows a faint band around ~25 kDa while pBSMu1+SpNprE+α-s2-casein has a band with a very high intensity at this size, corresponding to the size of SpNprE+α-s2-casein.
The western blot of the concentrated protein fraction containing proteins >30 kDa for SpNprE+α-s2-casein showed a band at ~25 kDa, while the wild type shows no band, indicating that the protein was secreted into the cell medium and was present in the concentrated fraction (Fig. 27).

With our protein of interest being ~27 kDa without and ~30 kDa with the signal peptide, it is highly possible that during the filter step, proteins with this particular size did not flow through the column, since the cutoff was not chosen properly.

Conclusion

α-s2-casein was successfully cloned and expressed in B. subtilis. With the focus on optimizing the expression of the protein of interest, five signal peptides (sslipA, SPNprE, SPPel, SPEpr and SPYurl) were tested out. No proteins were detected in the Coomassie stained growth media, indicating that further tests need to be carried out on concentrated fractions to get clearer results. Western Blot analysis of lysed pellet proteins illustrate no protein of interest in the pellet, but an unspecified signal that might be degraded or cleaved protein (Fig. 26). Western Blot analysis of concentrated media fraction containing proteins >30 kDa showed the protein of interest indicating that the α-s2-casein protein fused to the SPNprE signal peptide was secreted into the growth media (Fig. 27).
In comparison to pBSMUl1-SPNprE+α-s1-casein (BBa_K2924052) pBSMUl1-SPNprE+α-s2-casein (BBa_K2924052), where it appears that the protein of interest is retained in the cells, α-S2 casein seems to be efficiently secreted by the same signal peptide. This confirms our initial hypothesis that an optimal tag needs to be found for each individual protein. Many known B. subtilis secretion signals are available for that purpose.
By testing SPNprE, we were able to confirm it as a suitable secretion tag for α-s2-casein.

B. subtilis Hpall + RBS + SPYurl + α-s2-casein + 6xHis-tag + fd terminator

This composite part (Fig. 28) contains the constitutive promoter PHpall (BBa_K2924043), expressing α-s2-casein (BBa_K2924027) with the secretion signal SPYurl (BBa_K2924050) and the fd terminator (BBa_K2924044) for expression and secretion of the protein in B. subtilis.
Fig. 28: Scheme of expression construct for B. subtilis. The insert, containing the promoter PHpall (BBa_K2924043), α-s2-casein gene (BBa_K2924027) and the double fd terminator (BBa_K2924044), was cloned into the pBSMUl1 backbone with different secretion signals - here: SPYurl (K2924050 ).
For expression and secretion of α-s2-casein, the gene was cloned into the high-copy pBSMUl1-SPYurl plasmid, N-terminally fused to the signal peptide SPYurl and C-terminally fused to a 6xHis-tag for easier purification and immunodetection.

Due to time issues only the concentrated protein fraction containing proteins around the size and higher than 30 kDa was analysed by Western Blot(Fig. 29).
Fig. 29: Western Blot of B. subtilis concentrated media fraction containing proteins around and over 30 kDa. Wild type= B. subtilis DB430. The wild type shows a faint band around ~25 kDa while pBSMu1+SpYurl+α-s2-casein has a band with a increased intensity at this size, which fits to SpYurl+α-s2-casein.
The western blot of the concentrated media fraction containing proteins over and around 30 kDa indicates, that at least some of our target protein is secreted out of the B. subtilis cells, however not as strong as with another signal peptide (SPNprE).

Conclusion

α-s2-casein was successfully cloned and expressed in B. subtilis. With the focus on optimizing the expression of the gene of interest, five signal peptides (sslipA, SPNprE, SPPel, SPEpr and SPYurl) were tested out. Western Blot analysis of concentrated media fraction containing proteins >30 kDa shows the protein of interest, indicating successful secretion of α-s2-casein protein fused to the SPYurl signal peptide, albeit in low quantities, into the growth media (Fig. 29).

In comparison to pBSMUl1-SPNprE+α-s1-casein (BBa_K2924052) the secretion is increased, but in comparison to pBSMUl1-SPNprE+α-s2-casein (BBa_K2924052) the protein secretion is decreased. Many known B. subtilis secretion signals are available the purpose of individually optimizing secretion for each specific protein. With SpYurl, we found a second signal peptide next to SPNprE which is suitable for α-s2-casein secretion in B. subtilis.

Synechocystis

In the process of producing synthetic milk, we decided to also test protein production in a photoautotrophic organism. For this purpose, we engineered Synechocystis sp. PCC 6803 to express milk proteins using the strong constitutive promoter Pcpc560 (BBa_K2924000). First experiments were carried out with the fluorescent reporter gene mVenus (BBa_K2924035) to test the promoter strength.

Fig. 30: A: Fluorescence of the cultures. B: Fluorescence of the cultures normalized by optical density. Fluorescence was measured at 𝝺ex/em 512 /527 nm. Fluorescence was measured compared to the empty vector control to control for autofluorescence of the cells. Sterile BG11 was used as a blank to measure autofluorescence of the medium. Measurements were carried out in technical triplicates. Black: EVC. Yellow: Pcpc560:mVenus.
Fluorescence increased linearly with growth, while only slightly decreasing per cell with increasing optical density.
Fig. 31: SDS-PAGE of soluble fraction of synechocystis protein, extracted from either EVC or mVenus-expressing cultures after 2 days. The SDS-PAGE was run at 45 mA for 90 minutes and then stained with Coomassie blue.
We were able to detect the mVenus protein in an SDS-PAGE, showing a distinct band at 26.9 kDa. Furthermore, visualization of cells using confocal fluorescence microscopy revealed a strong fluorescence signal (Fig. 32).
Fig. 32: Fluorescence microscopy image of Synechocystis cells expressing mVenus. Autofluorescence of the chlorophyll is shown in magenta, fluorescence of mVenus is shown in green.
After proof-of-concept production of mVenus, we fused α-S1-casein to the same strong promoter.
Fig. 33: SDS-PAGE of total cyanobacterial protein, extracted from either EVC or α-s1-casein-expressing cultures after 2 days, separated into soluble and insoluble protein fractions. The gel was run at 220 V for 45 minutes and then stained with a Coomassie blue dye.
We were able to detect the α-s1-casein protein in an SDS-PAGE, showing a distinct band at 24.4 kDa. We also tested the growth behavior of the two protein production strains, shown in Fig. 34.
Fig. 34: Growth analysis of Synechocystis protein production strains compared to the empty vector control (EVC) in black. Gray: Pcpc560:α-s1-casein. Yellow: Pcpc560:mVenus.
Fig. 34 shows the growth behavior of the two Synechocystis production strains. In contrast to the empty vector control shown in black, both production strains grow slower. As discussed in our metabolic model, targeted production directly interferes with growth and biomass production. A decrease in growth in our production strain therefore indicates redirection of cellular resources towards the target protein.
Next to proof-of-concept production on a small scale, we also developed and tested protocols for larger scale production using the MINIFORS lambda photobioreactor. This enabled us to precisely control temperature, light, pH and gas flow. A special up-and-down agitation mechanism results in gentler mixing of the cells without sheering stress, as well as less foaming.

Using the photobioreactor, we grew our proof-of-concept strain expressing mVenus in standard BG11 media. To avoid excessive light stress from the LED-collar, we gradually increased the light intensity during the first few days of culturing.

Fig. 35: Growth behavior and protein yield from the Synechocystis Pcpc560 strain grown in the MINIFORS photobioreactor. A: Growth curve measured by OD750 over the course of 10 days. Optical density samples were taken in triplicates. B: Cell density-dependent total protein yield. Cells were grown in standard BG11 media using a gradual increase of light intensity from 10 to 50 µmol photons m-2 s-1, 1 % CO2, 5 Hz agitation and 30 ˚C.
As shown in Fig. 35 A, we were able to reach an OD750 of 12 before the culture reached stationary phase. The total protein yield measured (Fig. 35 B) is comparable with literature values obtained for E. coli 23, the most widely used model organism for protein production. Therefore, while there is still room for optimization, we propose that Synechocystis is a viable chassis for larger-scale protein production and should be further used by future generations of iGEM teams.

Pichia pastoris

The final milk protein we decided to produce is lactoferrin, which is catalytically active. Lactoferrin reduces inflammation factors and thus promotes the absorption of iron into the blood and its availability in the body. Lactoferrin is a multifunctional protein from the transferrin group, which is assigned to serine proteases due to its enzymatic activity. In addition, lactoferrin also has nuclease activity and is an iron chelator14.
The heterologous expression of proteins has an important industrial role, e.g. for the production of insulin17. Identification of new peptide and protein pharmaceuticals and the optimization of the expression of known pharmaceuticals represent a huge research sector. Around 155 pharmaceuticals and vaccines developed by bio-pharmaceutical companies were approved by the US Food and Drug Administration in 2002, the amount of approved recombinant proteins quickly rose to over 200 in 200917. Besides insulin, medically important proteins like albumin and the human growth hormone (HGH) are produced by microbes or higher organisms.
Small proteins are usually expressed in prokaryotic organisms like Escherichia coli, which enables easy, quick and cheap protein expression18. Disadvantages are the lack of post-translational modification and glycosylation, the difficult expression of large proteins and proteins with disulphide bonds18, 19.
These drawbacks can be compensated by using eukaryotic yeast as an expression host. Two of the primarily used yeast expression systems are Saccharomyces cerevisiae and Pichia pastoris20.

In P. pastoris, the AOX1 promoter is methanol inducible and therefore the AOX1 gene is highly expressed in the presence of methanol22. This leads to high recombinant protein yields of genes introduced into P. pastoris under influence of the AOX1 promoter23.
The pPICZB Vector from the EasySelect™ Pichia Expression Kit from Thermofisher Scientific was used, which contains an inducible AOX1 promoter and a Zeocin™ resistance gene.

The lactoferrin coding sequence originates from the organism Bos Taurus and was synthesized commercially. In addition, a NotI and an Eco72I interface were placed at both ends of the gene to clone the gene into the pPICZB vector (Fig. 36).

Fig. 36: The genetic organisation of the P. pastoris expression system with the AOX1 promoter (BBa_K2924039) followed by the target gene and C-terminally fused to a His-tag and AOX1 Transcription Terminator(BBa_K2924040). The vector is pPICZB. E: Eco72I. N: NotI
Results

After transformation of the construct shown in Fig. 36 and subsequent expression of lactoferrin in our P. pastoris strain, we investigated whether an increase in lactoferrin gene expression could be observed by performing RT-qPCR.
To measure the expression levels, P. pastoris RNA was isolated, cDNA was synthesised and as a negative control (RNA control), the same reaction was run without adding the reverse transcriptase. A negative control strain, containing only the empty vector, was also included. The qPCR was run with lactoferrin-specific qPCR primers. Data were normalized to the two P. pastoris housekeeping genes PpARG4 and PpTAF10.
Fig. 37 shows the relative expression of lactoferrin. While the EVC (blue) shows no expression at all, lactoferrin expression (red) is high.
Fig. 37: Relative gene expression level of Lactoferrin in P. pastoris 24 h after methanol induction. The expression level of each gene was normalised to the housekeeping genes PpARG4 and PpTAF10. The highest expressing strain of each construct was set to 100 % to gain relative expression levels.
While the RNA control also showed a signal, this is only minor and likely due to residual gDNA contamination.
In summary, we were able to show high expression of lactoferrin in P. pastoris based on qRT-PCR. This is the first step towards establishing P. pastoris as one of our milk protein production chassis. While there was no time to check protein amount and activity, we are confident that this organism is an excellent addition to our repertoire.

Final summary

To summarize all of our results, we were able to demonstrate that each of our engineered subsystems work.
We were able to produce various milk proteins in four completely different chassis - one gram-negative, one gram-positive, one eukaryotic and one photoautotrophic organism. Each of the six main milk proteins could be produced in E. coli. Furthermore, we were able to show successful secretion of our milk proteins to the media using B. subtilis. The catalytically active protein lactoferrin was successfully expressed in the yeast P. pastoris. Finally, we explored the photosynthetic organism Synechocystis as a protein production host and were able to show encouraging results, in both small- and large-scale.
In accordance with the promising protein results, we were also able to genetically modify Synechocystis sp. PCC 6803 for our fatty acid-specific goals. By overexpressing different thioesterases, we were able to show modular manipulation of fatty acid chain lengths. By CRISPRi-based gene repression, we were able to reduce unfavourable pathway reactions. Finally, we were able to improve an existing fatty acid biosensor, as well as introduce a completely new part which enables efficient in vivo measurements of fatty acids.
We thereby enabled synthesis of all building blocks needed to create synthetic cow’s milk using only Synechocystis.
We present a collection of methods, synthetic biology strategies, and parts specifically tailored to future applications using photosynthetic bacteria in the hope that future iGEM teams may benefit from our work. Finally, by establishing this strain as a main organism for producing our products, we present an eco-friendly production alternative because of its phototrophic abilities to reduce CO2 and increase O2 in the atmosphere.



References

  1. Kamran Jawed, Anu Jose Mattam, Zia Fatma, Saima Wajid, Malik Z. Abdin, Syed Shams Yazdani "Engineered production of short chain fatty acid in Escherichia coli using fatty acid synthesis pathway." PloS one 11.7 (2016): e0160035.
  2. Lun Yao, Ivana Cengic, Josefine Anfelt, Elton P. Hudson "Multiple gene repression in cyanobacteria using CRISPRi." ACS synthetic biology 5.3 (2015): 207-212.
  3. Jian-Hua Zhang, Poorni Adikaram, Mritunjay Pandey, Allison Genis, and William F. Simonds "Optimization of genome editing through CRISPR-Cas9 engineering." Bioengineered 7.3 (2016): 166-174.
  4. D. P. Clark and J. E. Cronan Jr., “Two-carbon compounds and fatty acids as carbon sources," in Escherichia coli and Salmonella: Cellular and Molecular Biology, F. C. Neidhardt, R.Curtiss III, J. L. Ingraham, et al., Eds., pp. 343–357, ASM Press,Washington, DC, USA, 1996
  5. Fuzhong Zhang, James M Carothers, Jay D Keasling. “Design of a dynamic sensor-regulator system for production of chemicals and fuels derived from fatty acids" Nature Biotechnology volume 30, pages 354–359 (2012)
  6. Mee-Jung Han, Jeong Wook Lee, Sang Yup Lee, and Jong Shin Yoo. “Proteome-Level Responses of Escherichia coli to Long-Chain Fatty Acids and Use of Fatty Acid Inducible Promoter in Protein Production" Hindawi Publishing Corporation Journal of Biomedicine and Biotechnology Volume 2008, Article ID 735101,12 pages
  7. Lee TS, Krupa RA, Zhang F, Hajimorad M, Holtz WJ, Prasad N, Lee SK, Keasling JD. "BglBrick vectors and datasheets: A synthetic biology platform for gene expression." J Biol Eng. 2011 Sep 20;5:12. 10.1186/1754-1611-5-12 PubMed 21933410
  8. Wood, W.B. 1966. Host specificity of DNA produced by Escherichia coli: bacterial mutations affecting the restriction and modification of DNA. J.Mol.Biol. 16:118-133
  9. Simonen, Marjo, and Ilkka Palva. "Protein secretion in Bacillus species." Microbiology and Molecular Biology Reviews 57.1 (1993): 109-137.
  10. Vincent Sewalt, Diane Shanahan, Lori Gregg, James La Marta and Roberto Carrillo "The Generally Recognized as Safe (GRAS) process for industrial microbial enzymes." Industrial Biotechnology 12.5 (2016): 295-302.
  11. IFBC (International Food Biotechnology Council). "Chapter 4: Safety Evaluation of Foods and Food Ingredients Derived from Microorganisms in Biotechnologies and Food: Assuring the Safety of Foods Produced by Genetic Modification." Reg. Toxicol. Pharmacol. 12 (1990): S1-S196.
  12. Food and Drug Administration. "Carbohydrase and protease enzyme preparations derived from Bacillus subtilis or Bacillus amyloliquefaciens: Affirmation of GRAS Status as direct food ingredients. 64 Fed." Reg. 19887 19895 (1999).
  13. Legrand, D., Elass, E., Carpentier, M., & Mazurier, J. (2005). Lactoferrin. Cellular and Molecular Life Sciences, 62(22), 2549.
  14. Ward, P. P., Paz, E., & Conneely, O. M. (2005). Lactoferrin. Cellular and Molecular Life Sciences, 62(22), 2540.
  15. Iyer, S., & Lonnerdal, B. (1993). Lactoferrin, lactoferrin receptors and iron metabolism. European journal of clinical nutrition.
  16. Ebina Y, Edery M, Ellis L, Standring D, Beaudoin J, Roth RA, Rutter WJ (1985) Expression of a functional human insulin receptor from a cloned cDNA in Chinese hamster ovary cells. Proc Natl Acad Sci U S A 82: 8014–8018
  17. Demain AL, Vaishnav P (2009) Production of recombinant proteins by microbes and higher organisms. Biotechnol Adv 27: 297–306
  18. Terpe K (2006) Overview of bacterial expression systems for heterologous protein production: From molecular and biochemical fundamentals to commercial systems. Appl Microbiol Biotechnol 72: 211–222
  19. Tran A-M, Nguyen T-T, Nguyen C-T, Huynh-Thi X-M, Nguyen C-T, Trinh M-T, Tran L-T, Cartwright SP, Bill RM, Tran-Van H (2017) Pichia pastoris versus Saccharomyces cerevisiae: a case study on the recombinant production of human granulocyte- macrophage colony-stimulating factor. BMC Res Notes 10: 148
  20. Cereghino JL, Cregg JM (2000) Heterologous protein expression in the methylotrophic yeast Pichia pastoris. FEMS Microbiol Rev 24: 45–66
  21. Byrne B (2015) Pichia pastoris as an expression host for membrane protein structural biology. Curr Opin Struct Biol 32: 9–17
  22. Lin-Cereghino J, Wong WW, Xiong S, Giang W, Luong LT, Vu J, Johnson SD, Lin- Cereghino GP (2005) Condensed protocol for competent cell preparation and transformation of the methylotrophic yeast Pichia pastoris. Biotechniques 38: 44–48
  23. New England Biolabs: https://international.neb.com/tools-and-resources/usage-guidelines/protein-data